Both IR and anti-Stokes Raman measurements on OH stretch vibrations nu(OH) of water see two time constants, similar to0.3 and similar to0.6 ps. We attribute the former to spectral diffusion and the latter to the nu(OH) lifetime. This interpretation is evident from our Raman spectra, which show the nu(OH) spectrum changing shape and the nu(OH) population decaying. The nu(OH) decay is accompanied by the rise (similar to0.6 ps) and subsequent fall (similar to1.4 ps) of population in the delta(OH) bending vibration. Bakker et al. [Chem. Phys. Lett. 385 (2004) 329] assert their IR data indicates an absence of (>100 fs) spectral diffusion, that the faster time constant is the nu(OH) lifetime and that the slower is the lifetime of a conjectured intermediate state. The properties of this conjectured intermediate differ significantly from the bending vibration directly observed by us. The intensity analysis by Bakker et al. said to contradict our interpretation relies entirely on ad hoc assumptions about excited-state absorptions. We show these ad hoc assumptions are unfounded and some are unreasonable, and that a plausible explanation for the IR intensities exists that is consistent with our interpretation. (C) 2003 Elsevier B.V. All rights reserved.